[OSM-talk-be] mapping farm-shop

Thu Apr 21 18:00:27 UTC 2016

Can we take a step back here? It’s starting to feel like we’re discussing
abandoned railways or something. Glenn, please don’t be the one to decide
who is the community and who is the einzelganger. This is defined by the
community, not by words but by actions. And please do not tell me to stop
doing something I never actually did. We are discussing an idea, no-one has
even touched an editor yet.

Don’t try to be the one to decide what is our core idea and what isn’t.
This is not how OpenStreetMap works: it started as something, then became
something else because mappers.

I’m going to stop quoting what you and I said, as we are clearly
misrepresenting each other’s words. You read several things into my words I
never meant – I’m hoping I’m not doing the same.

Some points I would like to make though.

* I’m not saying we SHOULD MAP shops in any special way. I am saying that
whenever the point “what you’re talking about shouldn’t be in OSM” is being
a made, I’ve never been quite convinced. There might be reasons to store
some information externally, but as Marc said that’s not straightforward at
all (I wrote a diary entry about it which got a lot of interesting feedback
[1], Ben Abelshausen and I have even worked on a rather big idea around the
subject).

* We’re talking about a general problem in the tagging scheme: often a
thing can be only one thing, while in reality, it is not. Say, a
hotel-restaurant. As Pieter said, solving this by using semicolons is often
discouraged, though it can make sense sometimes [2]. The wiki says on when
not to use: “in general avoid”. Hence Pieter’s quest for an alternative –
exactly as the wiki states.

* In the case of shops, we make huge amounts of categories to fit every
shop into something very specific. But reality isn’t that simple. To solve
this problem, there are some cases where it would be useful to know what
the shop sells. Just like the classification of “secondary” road doesn’t
tell me all I need to know – I wan’t to know the lanes, the width, the
maxweight, the road surface. We don’t try and make a definition of
“secondary” that is valid worldwide, because that makes no sense. But that
means that you need additional information to know what exactly it means,
especially as non-local users might use the data.

* Apparently, this is controversial when it comes to products sold in a
shop. Why is it controversial for a shop, but not for fuel stations,
vending machines, recycling containers, beers served at cafés (or is
openbeermap controversial?)?

* We’re talking about a concrete project with real community building
potential which would need a grand total of seven types of products. Have a
look at the website [3]

* The problem of updating is a general one. I do not trust the “you can’t
keep it up to date” argument, as it is the exact same argument used against
OpenStreetMap in general. “We have a dataset of roads, measured to the
centimetre. Why on earth would you use or make other data when ours is
available?” It might be a lot easier to keep addresses in an external
dataset, so why duplicate them in OSM? (I wrote about that in my diary
recently [4]).

In general, in these discussions I see people stating things like “we don’t
do that” (e.g. map opinions), and then when you say “well, we actually do”
(e.g. map smoothness), then they say: “well, that’s an exception”. If the
rules are exceptions, they’re not rules. They are guidelines to help you
judge, not laws that decide for you.

I think we should stay optimistic: we are still at the start of a curve.
Build it and they will come. The more complete the map, the more users we
will have. We would not have the huge amount of POIs added by maps.me
editors without a dataset good enough to allow for a userfriendly
navigation app. What looks crazy today, might just look obvious tomorrow.

If the plan would be to import a dataset of all Belgian shops with all
their products, I would totally be against. If someone says “hey, what’s
the best way to try this?”, then I will do my very best to think along.

Marc in fact automatically corrected all of those URLs and wrote a nice
diary post about it. Just an example of how URLs are not the most stable of
things either, nothing more. The point being that “stability” is a
spectrum, and I don’t know how to put a cut-off point of where a property
becomes “too unstable”, except relative to the community. A hypothetical
example: in the desert of Australia road smoothness evolves quickly. When
the truck comes to flatten the dirt, it’s super smooth. After a few weeks
of lorry travel, it’s all washboarded. It would be absolutely crazy to tray
and map this. Now imagine fifty crazy dudes (and one dudette, for good
measure) with sportscars, who are all waiting for the roads to get freshly
levelled. They get the contacts of the guys doing the flattening of the
roads and they update this information daily. Why not?

* “Nobody uses OpenStreetMap for that”: nobody used OSM for car navigation
before we had a road map. Nobody used it for hiking before we had a hiking
map. Nobody used it for navigating with a wheelchair before we had a
wheelmap. Again, not a valid argument in my opinion.

In the eyes of the general population, we are these fifty crazy dudes (and
one dudette).

1)
http://wiki.openstreetmap.org/wiki/Semi-colon_value_separator#When_NOT_to_use

2) http://www.openstreetmap.org/user/joost%20schouppe/diary/34328

3) http://www.boerenbrood.eu/

4) http://www.openstreetmap.org/user/joost%20schouppe/diary/38184
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-be/attachments/20160421/68a06eae/attachment.htm>